GenDigital

Reducing Notifications & Alerts Fatigue

Challenge

Improve the customer experience by reducing the volume and intrusiveness of Norton notifications, without compromising the revenue driven by cross-sell and upsell notifications.

My Impact

I played a central role throughout the exploration, design, and development phases. My focus was on making sense of large volumes of qualitative and quantitative data, translating insights into testable design concepts, and building scalable tools to help other teams do the same. I also developed a company-wide playbook and trained an AI assistant to support future implementation.

Role UX/UI design | Year 2025

Lloyd's Bank Mobile App

Process

Discovery
Understanding the data

Customer satisfaction with Norton 360 had been steadily declining. The data told a stark story driven in large part by over-messaging. Customers felt overwhelmed and under-informed. Many saw notifications as thinly veiled sales pitches. By January 2025, this issue alone had generated over 10,000 negative comments and pushed the NPS for messaging down to -62.

To fix it, we had to first understand the scale and nature of the problem.

We investigated multiple data sources. On average, customers were receiving 5.4 notifications per week, which is more than one every working day. Many were persistent, lingering on-screen until dismissed manually. Some were intrusive, full-screen takeovers that interrupted users mid-task.

We also discovered misleading UX:

  • “Don’t show me this again” often meant “hide for a few days,” not a true opt-out

  • Notification settings were inconsistent and often inaccessible

  • Some alerts couldn't be silenced at all

The story the data told was clear: the messaging experience wasn’t just noisy, it was broken.

Ideation
Fixing the Foundations

The challenges we need to tackle were three-fold:

  1. Over-notifying customers

  2. Broken notifications UX (using dark patterns for marketing and sales)

  3. Limited notification settings

To address this, we needed a structured way to differentiate meaningful alerts from noise. We built a classification framework and decision tree that divided notifications into five categories:

  • Critical

  • Warning

  • Informational

  • Success

  • Sales

For each category we devised a set of rules that would control when the notification was displayed, for how long it was displayed, what dismiss and snooze functionality was supported. We also created a set of controls for notification settings that link directly to each notification.

In addition, I created a comprehensive notifications component for the Design System that supported the differently display rules and functionality of the notifications in each of the categories listed above.

[Image of Design System Component]

Testing with Users
Validating with real feedback

We tested multiple notification types (Critical, Warning, Informational) across different designs and interactions. Key findings included:

  • Colour and iconography were vital for helping users assess urgency at a glance

  • Notification summaries after exiting fullscreen or focus mode were preferred to avoid disruption

  • Users valued the ability to “snooze” notifications, not just disable them entirely

  • Users valued more comprehensive notifications settings

    [notification settings image]

    [Notification summaries image]

Training Other Product Teams
Scaling across a global organisation

To make our notification overhaul scalable across a large, international product landscape, we needed a clear, reusable framework that any team could pick up and run with.

That’s why I created the Notifications Playbook: a comprehensive guide covering everything from deciding if a notification is necessary (using the NEED framework) to categorisation, timing, content guidelines, and governance. It gives designers, engineers, and PMs a shared language and decision-making toolkit.

But I didn’t stop at documentation.

We took it a step further and trained a GPT model on the playbook. Now, teams can query it in natural language to get instant, contextualised answers. For example, if someone asks:

“Should I show a toast when VPN turns off during fullscreen video?”

the assistant will walk them through the NEED framework, it will categorise a notification based on a security risk assessment, explain the best type of notification to use (push notification, in-app notification, email), and remind users of the display rules (e.g., self-dismiss times, contextual menu options) based on the security-level categorisation.

This approach turned our framework into a living, accessible assistant, which removes guesswork, speeds up decisions, and helps teams consistently deliver smarter, user-centred notifications.

The Playbook also detailed a new anatomy for notifications, according to it’s security-level category), helping designers keep CTAs, the more actions menu, colour, and iconography consistent over time and across teams.

[Image of Notifications Anatomy]

Outcomes
Fewer interruptions, smarter interations

Our changes delivered measurable impact across the board:

  • Critical notifications reduced by 43%
    We cut the number of alerts classified as "critical" from 24 to 16, ensuring users are only interrupted when absolutely necessary.

  • Shorter screen time for notifications
    Onscreen durations were drastically reduced:

    • Critical: from 20 minutes to 5 minutes max

    • Warning: 15 seconds

    • Informational and Success: just 5 seconds

  • No more fullscreen takeovers
    We eliminated intrusive dialogues completely. All notifications are now toast-style, keeping the experience lightweight and non-disruptive.

  • No more duplicate alerts for blocked websites
    Previously, users were hit with both a fullscreen dialog and an in-browser warning. We removed the redundancy—users now see a single, in-context notification.